CN104965184A - SPEN single-scanning magnetic resonance imaging spiral sampling and reconstructing method - Google Patents

Abstract

Provided is an SPEN (spatiotemporally-encoded) single-scanning magnetic resonance imaging spiral sampling and reconstructing method, relating to a magnetic resonance imaging method. 90 DEG linear scanning pulses and 180 DEG linear scanning pulses combine with corresponding SPEN gradients, and in a trigger period, protons in a space spin to obtain a quadratic phase relevant with a spatial position; in a sampling period, a spiral sampling gradient after optimization collets data, and obtains spatial domain magnetic resonance imaging data with T2*weighting in a superfast speed; and finally, spiral sampling data are reconstructed through a specific grid algorithm and a super-resolution reconstructing method based on compression perception, so as to obtain super-resolution high quality magnetic resonance images. The method greatly improves the image quality of SPEN single-scanning imaging, and provides a great imaging tool for the fields requiring superfast imaging.

Description

Based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep

Technical field

The present invention relates to MR imaging method, especially relate to a kind of spiral sampling based on the magnetic resonance imaging of space-time code single sweep and method for reconstructing.

Background technology

In magnetic resonance imaging, the resolution improving image taking speed or interior raising imaging is at a fixed time that people seek assiduously always.Ultra-fast imaging techniques needs to play an important role in high time-resolved experiment at some, as functional imaging (functional MRI, fMRI) ^[1-4], freely breathe cardiac imaging (free-breath heart imaging) ^[5,6]with the experiment of some higher-dimensions as diffusion tensor imaging (diffusion tensor imaging, DTI) ^[7-9].In numerous supper-fast methods, the Echo-plane imaging (echo planar imaging, EPI) of single sweep becomes most popular ultra-fast imaging techniques by means of its high temporal resolution ^[10-12].As everyone knows, Echo-plane imaging, when once exciting, is sampled by a series of echo train gradient, just can be filled up whole k-space.But this sample mode is easy to be subject to nonuniform field impact, causes last image to occur distortion ^[13,14].In order to overcome echo planar imaging to non-uniform magnetic field sensitive issue, the Frydman group of Wiesmann research institute of Israel proposes the space-time code (spatiotemporally-encoded of single sweep, SPEN) MR imaging method, the image-forming principle of this method and Echo-plane imaging is essentially different ^[15].For Echo-plane imaging, the signal obtained in a certain sampling instant is the contribution being derived from all proton spins in space.And space-time code method is only derived from the contribution of a certain ad-hoc location proton spin in space at the signal that a certain sampling instant obtains.Phase encoding " bilp " gradient in Echo-plane imaging sequence is just replaced by this supper-fast space-time code method, but remain the sampled gradients of echo planar imaging sequence, therefore this space-time code method has the image taking speed identical with echo planar imaging.In a certain sampling instant, the intensity of space-time code imaging signal only depends on the local spin densities in corresponding locus.This characteristic is given space-time code MR imaging method and is resisted B to a certain extent ₀the uneven sex ability in field.Space-time code method ensure that space-time code image for the robustness of nonuniform field, compares echo planar imaging image, and it is uneven and stand larger B to resist local susceptibility ₀uneven field ^[16-18].

Due to the characteristic that it is excellent, in the past few years, the supper-fast imaging of space-time code is developed rapidly.But, still there is following defect in the supper-fast formation method of current space-time code single sweep: first, adopt the positive and negative gtadient echo chain switched fast to carry out data acquisition in sample phase, require higher to hardware device, and its image is easy to the impact by eddy effect; ^[19,20]the second, due to T ₂the impact of relaxation, sampling number can not be too much, which has limited the inherent spatial resolution of image; ^[21]3rd, because counting of the full sampling request of the supper-fast imaging of space-time code is comparatively large, common sampling number can not meet Nyquist Sampling Theorem, thus occurs aliasing artefacts in reconstruction image ^[18].

As can be seen here, develop a kind of magnetic resonance image (MRI) obtaining high-quality in single sweep situation of can putting forward to be necessary.

List of references:

[1]Ciobanu L,Solomon E,Pyatigorskaya N,Roussel T,Le Bihan D,Frydman L.fMRI contrast at high and ultrahigh magnetic fields:Insight from complementarymethods[J].NeuroImage 2015；113(0):37-43.

[2]Moonen C,van Zijl P,Frank JA,Le Bihan D,Becker ED.Functional magneticresonance imaging in medicine and physiology[J].Science 1990；250(4977):53-61.

[3]Chang C,Glover GH.Variable-density spiral-in/out functional magneticresonance imaging[J].Magn Reson Med 2011；65(5):1287-1296.

[4]Ben-Eliezer N,Goerke U,Ugurbil K,Frydman L.Functional MRI usingsuper-resolved spatiotemporal encoding[J].Magn Reson Imaging2012；30(10):1401-1408.

[5]Frindel C,Robini M,Croisille P,Zhu YM.Comparison of regularizationmethods for human cardiac diffusion tensor MRI[J].Med Image Anal 2009；13(3):405-418.

[6]Bo Z,Haldar JP,Brinegar C,Zhi-Pei L.Low rank matrix recovery forreal-time cardiac MRI.201014-17April 2010.p 996-999.

[7]Ellingson BM,Sulaiman O,Kurpad SN.High-resolution in vivo diffusiontensor imaging of the injured cat spinal cord using self-navigated,interleaved,variable-density spiral acquisition(SNAILS-DTI)[J].Magn Reson Imaging2010；28(9):1353-1360.

[8]Scherrer B,Gholipour A,Warfield SK.Super-resolution reconstruction toincrease the spatial resolution of diffusion weighted images from orthogonalanisotropic acquisitions[J].Med Image Anal 2012；16(7):1465-1476.

[9]Frank LR,Jung Y,Inati S,Tyszka JM,Wong EC.High efficiency,lowdistortion 3D diffusion tensor imaging with variable density spiral fast spin echoes(3D DW VDS RARE)[J].Neuroimage 2010；49(2):1510-1523.

[10]Ahn CB,Kim JH,Cho ZH.High-speed spiral-scan echo planar NMR imaging[J].IEEE Trans Med Imaging 1986；5(1):2-7.

[11]Stehling MK,Turner R,Mansfield P.Echo-planar imaging:Magneticresonance imaging in a fraction of a second[J].Science 1991；254(5028):43-50.

[12]Cohen MS,Schmitt F.Echo planar imaging before and after fMRI:A personalhistory[J].Neuroimage 2012；62(2):652-659.

[13]Zeng H,Constable RT.Image distortion correction in EPI:Comparison offield mapping with point spread function mapping[J].Magn Reson Med2002；48(1):137-146.

[14]Chen NK,Avram AV,Song AW.Two-dimensional phase cycled reconstructionfor inherent correction of echo-planar imaging Nyquist artifacts[J].Magn Reson Med2011；66(4):1057-1066.

[15]Tal A,Frydman L.Single-scan multidimensional magnetic resonance[J].Prog Nucl Magn Reson Spectrosc 2010；57(3):241-292.

[16]Ben-Eliezer N,Irani M,Frydman L.Super-resolved spatially encodedsingle-scan 2D MRI[J].Magn Reson Med 2010；63(6):1594-1600.

[17]Chen L,Bao LJ,Li J,Cai SH,Cai CB,Chen Z.An aliasing artifacts reducingapproach with random undersampling for spatiotemporally encoded single-shot MRI[J].J Magn Reson 2013；237(0):115-124.

[18]Cai CB,Dong JY,Cai SH,Li J,Chen Y,Bao LJ,Chen Z.An efficientde-convolution reconstruction method for spatiotemporal-encoding single-scan 2DMRI[J].J Magn Reson 2013；228:136-147.

[19]Truong TK,Chen NK,Song AW.Application of k-Space Energy SpectrumAnalysis for Inherent and Dynamic B(0)Mapping and Deblurring in Spiral Imaging[J].Magn Reson Med 2010；64(4):1121-1127.

[20]Aldefeld B,Bornert P.Effects of gradient anisotropy in MRI[J].Magn ResonMed 1998；39(4):606-614.

[21]Chen L,Li J,Zhang M,Cai SH,Zhang T,Cai CB,Chen Z.Super-resolvedenhancing and edge deghosting(SEED)for spatiotemporally encoded single-shot MRI[J].Med Image Anal 2015；23(1):1-14.

Summary of the invention

The object of the present invention is to provide and do not need quick switch sampling gradient, very big reduction to the requirement of hardware, and can also improve the spatial resolution of image and resist a kind of spiral sampling based on the magnetic resonance imaging of space-time code single sweep and the method for reconstructing that lack sampling causes aliasing artefacts.

The present invention includes following steps:

1) on magnetic resonance imager operator's console, open the corresponding function software of magnetic resonance imager, first area-of-interest location is carried out to imaging object, then carry out tuning, automatic shimming, power and frequency correction;

2) measure the power of 90 ° and 180 ° linear frequency sweep pulses (chirp pulse) successively by the pulse train of measure linear scanning frequency pulse power, record survey the power of 90 ° and 180 ° linear frequency sweep pulses;

3) in NMR imaging instrument, bidimensional space-time code single sweep sequence compiled is in advance imported; Open the bidimensional space-time coding block of sequence, call in required 90 ° and 180 ° of linear frequency sweep pulses (chirp pulse), by measured performance number assignment in corresponding power and variable;

4) correlation parameter of chirp pulse and visual field is imported, and the maximum amplitude of sampled gradients and switching rate are set, spiral sampling gradient file is produced by spiral gradient generator program compiled in advance, spiral sampling gradient file is imported in sequence, and corresponding sampled gradients value is set;

5) step 4 is performed) the bidimensional space-time code single sweep sequence that sets, carry out data sampling, after data sampling completes, perform next step;

6) after data acquisition completes, first calculate stable phase angle point corresponding to sampling instant according to sequential parameter, secondly according to the smoothing process of the phase place of stable phase angle point to sampled signal, make the phase place change of data slowly;

7) to step 6) signal that obtains carries out signal gridding process, and the signal of helical trajectory is converted into the signal under Cartesian coordinates;

8) to step 7) signal that obtains carries out the super-resolution rebuilding of bidimensional space-time code, obtains the magnetic resonance image (MRI) of high-quality.

In step 2) in, the pulse train of described measure linear scanning frequency pulse power is the space-time code sequence of an one dimension, be made up of the combination of a linear frequency sweep pulse and space-time code gradient and sampled gradients, wherein sampled gradients and encode gradient act on same one dimension, gradient area equation, direction is contrary.Carry out many experiments by arranging different linear frequency sweep pulse power value, thus find suitable linear frequency sweep pulse power value.

In step 3) in, the structure of described bidimensional space-time coding block is followed successively by: 90 ° of linear frequency sweep pulses, 180 ° of linear frequency sweep pulses, sinc pulse, the PE direction of 180 ° are biased gradient;

Described 90 ° of linear frequency sweep pulses are in conjunction with spatial encoding gradient G _pespace encoding is carried out to low bandwidth dimension (PE direction);

Described 180 ° of linear frequency sweep pulses are in conjunction with spatial encoding gradient G _respace encoding is carried out to high bandwidth dimension (RO direction);

Described 180 ° of sinc pulse binding layers select gradient G _sscarry out layer choosing;

The described area being biased gradient in PE direction is spatial encoding gradient G _pethe half of area, positive and negative in spatial encoding gradient G _peon the contrary.

In step 4) in, described spiral sampling gradient file is made up of the gradient chain acting on RO, PE direction respectively, gradient file on RO and PE direction is undertaken calculating and generating by the optimum procedure write in advance, be made up of the positive and negative switched gradients of a series of slow change, gradient magnitude in both direction is no more than the greatest gradient amplitude that instrument specifies, switching rate is no more than the maximum switching rate of instrument regulation; For guaranteeing to decode to the whole visual field, the maximum value of PE direction sampled gradients cumulative area equals G _pethe half of area, the maximum value of RO direction sampled gradients cumulative area equals G _rearea.

In step 6) in, the computing formula of described stable phase angle point can be:

$\left\{\begin{array}{c}x\left(n\right)=\frac{{\mathrm{\Σ}}_1^N{G}_{\mathrm{ra}}\left(n\right)\mathrm{\τ}}{{2G}_{\mathrm{re}}{T}_{\mathrm{re}}}{L}_x,n=\mathrm{1,2},...,N\\ y\left(n\right)=\frac{{\mathrm{\Σ}}_1^N{G}_{\mathrm{pa}}\left(n\right)\mathrm{\τ}}{G_{\mathrm{pe}}{T}_{\mathrm{pe}}}{L}_y,n=\mathrm{1,2},...,N\end{array}\right.$

Wherein, N is sampling number, and τ is sampling interval duration, G _raand G _pafrequency encoding gradient and phase encoding gradient respectively, G _reand G _pefrequency decoding gradient and phase decoding gradient respectively, T _reand T _refrequency decoding gradient duration and phase decoding gradient duration respectively, L _xand L _ythe visual field in frequency coding direction and the visual field of phase-encoding direction respectively.

The formula of described smoothing processing can be:

$S^{\′}=S\·\exp [i\·\mathrm{\γ}(\frac{G_{\mathrm{pe}}{T}_{\mathrm{pe}}{y}^2}{2L_y}-\frac{G_{\mathrm{re}}{T}_{\mathrm{re}}{x}^2}{L_x})]$

Wherein, S is the signal that obtains of sampling, and γ is magnetogyric ratio, S ' be smoothing processing after data.

In step 7) in, described signal gridding process is by carrying out space interpolation to realize to the signal after level and smooth, except the signal after level and smooth in Interpolation Process, also needs to use luv space location distribution information and interpolation space location distribution information; Luv space location distribution information is step 6) in stable phase angle point x and y that calculate, the spatial positional information under the corresponding Cartesian coordinates of interpolation space location distribution information.Under normal circumstances, the interval between the locus under Cartesian coordinates can obtain less, thus is reduced in the error introduced in gridding process.

In step 8) in, the super-resolution rebuilding of described bidimensional space-time code is based on compressed sensing algorithm, and its ultimate principle utilizes image to be that this priori of rarefaction representation removes aliasing artefacts or improves spatial resolution inside domain of variation;

The formula of described super-resolution rebuilding is:

$\arg \min {\left|\right|\mathrm{\Φ\ρ}-S^{\′}\left|\right|}_2^2+\mathrm{\λ}{\left|\right|\mathrm{\ψ\ρ}\left|\right|}_1$

Wherein, Φ is the bidimensional space-time code perception matrix that calculates of parameter by experiment, and ρ is the high-quality magnetic resonance image (MRI) of rebuilding out, S ' be smoothing processing after data, λ is the weighting factor of sparse transformation and fidelity item, and Ψ is sparse transformation matrices.Solve formula above by iterative algorithm and just can obtain high resolving power, high-quality magnetic resonance image (MRI) without aliasing artefacts.

As preferably, described spiral sampling gradient needs to carry out corresponding optimization according to different hardware conditions and experiment article profile, enables its greatest gradient amplitude and greatest gradient switching rate meet hardware requirement, and can improve the sampling density of signal area.First described method for reconstructing is to the smoothing process of signal obtained of sampling, and makes the phase place change of data slowly; Then gridding is carried out to the signal after level and smooth, finally super-resolution rebuilding is carried out to the data of gridding.

First the present invention proposes the spiral sampling of bidimensional space-time code, and compared with sampling with conventional Cartesian, spiral sampling can reduce hsrdware requirements, improves the sampling density of signal area; Then introduce gridding and the super-resolution rebuilding based on compressed sensing, obtain the magnetic resonance image (MRI) of high-quality.

Spiral sampling based on the magnetic resonance imaging of bidimensional space-time code single sweep provided by the invention and method for reconstructing can overcome Descartes's Gradient of sampling and switch the eddy effect and higher hardware requirement brought fast, can when not increasing the sampling time, improve the sampling density of signal area, thus improve the spatial resolution of image.And the supper-fast imaging of the space-time code based on spiral sampling can overcome Descartes and to sample down the lack sampling aliasing artefacts brought, in conjunction with specific GRIDDING WITH WEIGHTED AVERAGE and the super-resolution rebuilding algorithm based on compressed sensing, high resolving power, high-quality magnetic resonance image (MRI) without aliasing artefacts can be obtained.Compare the supper-fast formation method of Descartes's space-time code, method efficiency of the present invention is better, imaging effect is better.

Accompanying drawing explanation

Fig. 1 is bidimensional space-time code single sweep spiral sequence of the present invention, and parameters title all provides in detail in sequence chart.

Fig. 2 is the flow process of bidimensional space-time code single sweep spiral sampling and reconstruction:

A contoured profile figure that () is object;

B () is the helical trajectory figure of design;

C spiral sampling gradient map that () is Program Generating;

D spiral data that () arrives for sequence acquisition;

E () is the experimental data after smoothing processing;

F () is the experimental data after gridding;

G () is the magnetic resonance image (MRI) after super-resolution rebuilding.

Fig. 3 is the rat brain magnetic resonance image (MRI) under different sequence equivalent environment:

A reference figure that () obtains for many scanning sequences;

B image that () obtains for Echo-plane imaging;

C image that () obtains for the supper-fast imaging of Descartes's space-time code;

D image that () obtains for space-time code spiral sampling.

Embodiment

Below in conjunction with the drawings and specific embodiments, the present invention will be further described.

The invention provides the spiral sampling based on the magnetic resonance imaging of bidimensional space-time code single sweep and method for reconstructing, each step in specific implementation process is as follows:

(1) first area-of-interest location is carried out to imaging object, then carry out tuning, automatic shimming, power and frequency correction.

(2) measure successively by the pulse train of measure linear scanning frequency pulse power and record the power of 90 ° and 180 ° linear frequency sweep pulses (chirp pulse).

Preferred use one dimension space-time code sequence pair power and variable carries out array parameter measurement, and object guarantees the loss that signal is the least possible.

(3) in NMR imaging instrument, single sweep bidimensional space-time code sequence compiled is in advance imported; And call in required 90 ° and 180 ° of linear frequency sweep pulses (chirp pulse), by the performance number assignment of 90 ° and 180 ° linear frequency sweep pulses (chirp pulse) measured in step (2) in corresponding power and variable.

The structure of described single sweep bidimensional space-time coding block is followed successively by: 90 ° of linear frequency sweep pulses, 180 ° of linear frequency sweep pulses, sinc pulse, the PE direction of 180 ° are biased gradient.

Described 90 ° of linear frequency sweep pulses are in conjunction with spatial encoding gradient G _pespace encoding is carried out to low bandwidth dimension (PE direction); Described 180 ° of linear frequency sweep pulses are in conjunction with spatial encoding gradient G _respace encoding is carried out to high bandwidth dimension (RO direction); Described 180 ° of sinc pulses and layer select gradient G _sscarry out layer choosing; And then on PE direction, apply a biased gradient, its gradient area is described spatial encoding gradient G _pethe half of area, positive and negative and described spatial encoding gradient G _peon the contrary.

The step arranging the parameters of described bidimensional space-time code single sweep sequence is: each correlation module first opening described single sweep bidimensional space-time code sequence, comprises bidimensional space-time coding block, echo time delay module.Then arrange according to area-of-interest size and carry out corresponding experiment parameter, comprise the stimulating frequency width Delta Ο of 90 ° of linear frequency sweep pulses ₉₀, firing time T ₉₀, power tpwr ₉₀, the stimulating frequency width Delta Ο of 180 ° of linear frequency sweep pulses ₁₈₀, firing time T ₁₈₀, power tpwr ₁₈₀, the visual field L in frequency coding direction _x, the visual field L of phase-encoding direction _y, imaging layer thickness thk, destroys intensity and the time of gradient.

(4) generate spiral sampling gradient file, spiral sampling gradient file is imported in sequence, and corresponding sampled gradients value is set.

Described generation spiral sampling gradient file refers to and imports to by spiral gradient generator program compiled in advance generation spiral sampling gradient file by relevant informations such as chirp pulse and visual fields, and the maximum amplitude of sampled gradients and switching rate are set, thus generate the spiral sampling gradient file meeting hsrdware requirements.

(5) the bidimensional space-time code single sweep spiral imaging sequence set is performed

Be different from conventional many scanning imageries sequence, this method once can excite and obtain view picture magnetic resonance image (MRI), thus can avoid motion artifacts; What be different from conventional single sweep sequence is that this method can be carried out the low coverage imaging of two dimension and folding artifact can not occur, and can improve the resolution of image or accelerate image taking speed further.And compare conventional single sweep formation method, this method has better performance on nonuniform field.If carry out multilayer imaging or functional imaging, first sequence will postpone one RD period, and object is to allow magnetization vector relaxation recover; Then, the various piece of pulse train acts on imaging object successively, and namely bidimensional space-time code, layer choosing, time-lag action are until data sampling terminates, and finally obtain the space-time code magnetic resonance image data of two dimension.Obtain time about tens μ s to hundreds of μ s of a width magnetic resonance image (MRI).

(6) data sampling will to the smoothing process of signal after completing, and first will calculate stable phase angle point corresponding to different sampling instant according to sampling parameter, the formula of stable phase angle point is:

$\left\{\begin{array}{c}x\left(n\right)=\frac{{\mathrm{\Σ}}_1^N{G}_{\mathrm{ra}}\left(n\right)\mathrm{\τ}}{{2G}_{\mathrm{re}}{T}_{\mathrm{re}}}{L}_x,n=\mathrm{1,2},...,N\\ y\left(n\right)=\frac{{\mathrm{\Σ}}_1^N{G}_{\mathrm{pa}}\left(n\right)\mathrm{\τ}}{G_{\mathrm{pe}}{T}_{\mathrm{pe}}}{L}_y,n=\mathrm{1,2},...,N\end{array}\right.$

Wherein N is sampling number, and τ is sampling interval duration, G _raand G _pafrequency encoding gradient and phase encoding gradient respectively, G _reand G _pefrequency decoding gradient and phase decoding gradient respectively, T _reand T _refrequency decoding gradient duration and phase decoding gradient duration respectively, L _xand L _ythe visual field in frequency coding direction and the visual field of phase-encoding direction respectively.Then apply an extra phase place to sampled signal, be used for eliminating the phase place concuss that quadratic phase produces, its formula is:

$S^{\′}=S\·\exp [i\·\mathrm{\γ}(\frac{G_{\mathrm{pe}}{T}_{\mathrm{pe}}{y}^2}{2L_y}-\frac{G_{\mathrm{re}}{T}_{\mathrm{re}}{x}^2}{L_x})]$

Wherein S is the signal that obtains of sampling, and γ is magnetogyric ratio, S ' be smoothing processing after data.Sampled data after smoothing processing, its PHASE DISTRIBUTION slowly, facilitates ensuing gridding operation.

(7) gridding process is carried out to sampled data.Gridding process need to use obtain in previous step level and smooth after data and the information of stable phase angle point, in addition, we also need to produce the locus distributed intelligence under Cartesian coordinates.The principle of gridding process is exactly by interpolation operation, is the data under Cartesian coordinates, facilitates ensuing super-resolution rebuilding by the script data transformations be distributed on helical trajectory.

(8) super-resolution rebuilding is carried out to the data after gridding.The super-resolution rebuilding of bidimensional space-time code is based on compressed sensing algorithm, its ultimate principle utilizes image to be that this priori of rarefaction representation removes aliasing artefacts or improves spatial resolution inside domain of variation, and the formula of the super-resolution rebuilding algorithm used in the present invention is:

$\arg \min {\left|\right|\mathrm{\Φ\ρ}-S^{\′}\left|\right|}_2^2+\mathrm{\λ}{\left|\right|\mathrm{\ψ\ρ}\left|\right|}_1$

Wherein Φ is the bidimensional space-time code perception matrix that calculates of parameter by experiment, and ρ is the high-quality magnetic resonance image (MRI) of rebuilding out, S ' be smoothing processing after data, λ is the weighting factor of sparse transformation and fidelity item, and Ψ is sparse transformation matrices.Solve formula above by iterative algorithm and just can obtain high resolving power, high-quality magnetic resonance image (MRI) without aliasing artefacts.

Below provide specific embodiment:

By bidimensional space-time code single sweep spiral sampling and method for reconstructing thereof, live body SD rat carries out embodiment displaying, be used for verifying feasibility of the present invention.Experiment test carries out on a Varian 7T imager (Agilent Technologies, SantaClara, CA, USA).Test the live body SD rat that the imaging sample adopted is about 250g, first use the chloral hydrate solution of 10% before experiment, in 0.4mg/100g ratio, injecting anesthetic is carried out to rat, test accordingly after waiting rat to enter dormant state.Before testing, first rat is fixed on experimental bed, then sends in imager.On magnetic resonance imager operator's console, open the corresponding function software of imager, interested rat position is positioned, select to carry out axial surface imaging to the brain of rat herein.After imaging region is had good positioning, carry out tuning, automatic shimming, frequency and capability correction.In order to evaluate the validity of spiral sampling method, under equivalent environment, Echo-plane imaging and traditional supper-fast imaging experiment of Descartes's space-time code are carried out as a comparison.According to the operating process of above-mentioned bidimensional space-time code single sweep spiral sampling and method for reconstructing thereof, before carrying out single sweep bidimensional space-time code, first measure the power of 90 ° and 180 ° used linear frequency sweep pulses respectively by the space-time code sequence of one dimension.Then compiled (as shown in Figure 1) bidimensional space-time code single sweep sequence is imported, open each correlation module of pulse train, comprise bidimensional space-time coding block and decoding sampling module, experiment parameter is set, specifically for the sample that the present embodiment adopts, its experiment optimum configurations is as follows: the stimulating frequency width Delta Ο of 90 ° of linear frequency sweep pulses ₉₀for 64kHz, firing time T ₉₀be the stimulating frequency width Delta Ο of 3 μ s, 180 ° of linear frequency sweep pulses ₁₈₀for 32kHz, firing time T ₁₈₀be 3 μ s, total sampling number is the visual field L in 6996, RO direction _xfor the visual field L in 4.5cm, PE direction _yfor 4.5cm, imaging layer thickness thk is 2mm, and the intensity destroying gradient is 3.0Gs/cm, and the time is 1 μ s.After being set by above experiment parameter, the sampling time directly running whole sequence is about 60 μ s.

After data sampling is complete, rebuild spiral sampling data according to above-mentioned steps (6) ~ (8), net result as shown in Figure 3 D.As can be seen from Figure 3, under identical magnetic field environment, the distortion of echo planar imaging image (Fig. 3 B) is comparatively serious, and space-time code image (Fig. 3 C and D) distortion is less.Image (Fig. 3 C) relatively under Descartes's sample track and the image (Fig. 3 D) under spiral sampling track can be found out, spiral sampling can obtain the rat brain information that spatial resolution is better, details is enriched more.As can be seen from the region of yellow arrows indication, the image under Descartes's sample track can be subject to the impact of aliasing artefacts, thus occurs extra signal, and in image under spiral sampling track, aliasing artefacts does not occur.

In sum, spiral sampling based on the magnetic resonance imaging of bidimensional space-time code single sweep proposed by the invention and method for reconstructing effectively can overcome the aliasing artefacts that lack sampling causes, and improve the spatial resolution of image, high resolving power, magnetic resonance image (MRI) without aliasing artefacts can be obtained in single sweep situation.

Claims (9)

1., based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that comprising the following steps:

1) on magnetic resonance imager operator's console, open the corresponding function software of magnetic resonance imager, first area-of-interest location is carried out to imaging object, then carry out tuning, automatic shimming, power and frequency correction;

2) measure the power of 90 ° and 180 ° linear frequency sweep pulses (chirp pulse) successively by the pulse train of measure linear scanning frequency pulse power, record survey the power of 90 ° and 180 ° linear frequency sweep pulses;

3) in NMR imaging instrument, bidimensional space-time code single sweep sequence compiled is in advance imported; Open the bidimensional space-time coding block of sequence, call in required 90 ° and 180 ° of linear frequency sweep pulses (chirp pulse), by measured performance number assignment in corresponding power and variable;

4) correlation parameter of chirp pulse and visual field is imported, and the maximum amplitude of sampled gradients and switching rate are set, spiral sampling gradient file is produced by spiral gradient generator program compiled in advance, spiral sampling gradient file is imported in sequence, and corresponding sampled gradients value is set;

5) step 4 is performed) the bidimensional space-time code single sweep sequence that sets, carry out data sampling, after data sampling completes, perform next step;

6) after data acquisition completes, first calculate stable phase angle point corresponding to sampling instant according to sequential parameter, secondly according to the smoothing process of the phase place of stable phase angle point to sampled signal, make the phase place change of data slowly;

7) to step 6) signal that obtains carries out signal gridding process, and the signal of helical trajectory is converted into the signal under Cartesian coordinates;

8) to step 7) signal that obtains carries out the super-resolution rebuilding of bidimensional space-time code, obtains the magnetic resonance image (MRI) of high-quality.

2. as claimed in claim 1 based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that in step 2) in, the pulse train of described measure linear scanning frequency pulse power is the space-time code sequence of an one dimension, be made up of the combination of a linear frequency sweep pulse and space-time code gradient and sampled gradients, wherein sampled gradients and encode gradient act on same one dimension, gradient area equation, direction is contrary.

3. as claimed in claim 1 based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that in step 3) in, the structure of described bidimensional space-time coding block is followed successively by: 90 ° of linear frequency sweep pulses, 180 ° of linear frequency sweep pulses, sinc pulse, the PE direction of 180 ° are biased gradient;

Described 90 ° of linear frequency sweep pulses are in conjunction with spatial encoding gradient G _pespace encoding is carried out to low bandwidth dimension and PE direction;

Described 180 ° of linear frequency sweep pulses are in conjunction with spatial encoding gradient G _respace encoding is carried out to high bandwidth dimension and RO direction;

Described 180 ° of sinc pulse binding layers select gradient G _sscarry out layer choosing;

The described area being biased gradient in PE direction is spatial encoding gradient G _pethe half of area, positive and negative in spatial encoding gradient G _peon the contrary.

4. as claimed in claim 1 based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that in step 4) in, described spiral sampling gradient file is made up of the gradient chain acting on RO, PE direction respectively, gradient file on RO and PE direction is undertaken calculating and generating by the optimum procedure write in advance, be made up of the positive and negative switched gradients of a series of slow change, gradient magnitude in both direction is no more than the greatest gradient amplitude that instrument specifies, switching rate is no more than the maximum switching rate of instrument regulation; For guaranteeing to decode to the whole visual field, the maximum value of PE direction sampled gradients cumulative area equals G _pethe half of area, the maximum value of RO direction sampled gradients cumulative area equals G _rearea.

5., as claimed in claim 1 based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that in step 6) in, the computing formula of described stable phase angle point is:

$\left\{\begin{array}{cc}x\left(n\right)=\frac{{\mathrm{\Σ}}_1^N{G}_{\mathrm{ra}}\left(n\right)\mathrm{\τ}}{2G_{\mathrm{re}}{T}_{\mathrm{re}}}{L}_x,& n=\mathrm{1,2},...,N\\ y\left(n\right)=\frac{{\mathrm{\Σ}}_1^N{G}_{\mathrm{pa}}\left(n\right)\mathrm{\τ}}{G_{\mathrm{pe}}{T}_{\mathrm{pe}}}{L}_y,& n=\mathrm{1,2},...,N\end{array}\right.$

Wherein, N is sampling number, and τ is sampling interval duration, G _raand G _pafrequency encoding gradient and phase encoding gradient respectively, G _reand G _pefrequency decoding gradient and phase decoding gradient respectively, T _reand T _refrequency decoding gradient duration and phase decoding gradient duration respectively, L _xand L _ythe visual field in frequency coding direction and the visual field of phase-encoding direction respectively.

6., as claimed in claim 1 based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that in step 6) in, the formula of described smoothing processing is:

$S^{\′}=S\·\exp [i\·\mathrm{\γ}(\frac{G_{\mathrm{pe}}{T}_{\mathrm{pe}}{y}^2}{2L_y}-\frac{G_{\mathrm{re}}{T}_{\mathrm{re}}{x}^2}{L_x})]$

Wherein, S is the signal that obtains of sampling, and γ is magnetogyric ratio, S ' be smoothing processing after data.

7. as claimed in claim 1 based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that in step 7) in, described signal gridding process is by carrying out space interpolation to realize to the signal after level and smooth, except the signal after level and smooth in Interpolation Process, also need to use luv space location distribution information and interpolation space location distribution information; Luv space location distribution information is step 6) in stable phase angle point x and y that calculate, the spatial positional information under the corresponding Cartesian coordinates of interpolation space location distribution information.

8. as claimed in claim 1 based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that in step 8) in, the super-resolution rebuilding of described bidimensional space-time code is based on compressed sensing algorithm, and its ultimate principle utilizes image to be that this priori of rarefaction representation removes aliasing artefacts or improves spatial resolution inside domain of variation;

The formula of described super-resolution rebuilding is:

$\arg {\min \left|\right|\mathrm{\Φ\ρ}-S^{\′}\left|\right|}_2^2+\mathrm{\λ}{\left|\right|\mathrm{\Ψ\ρ}\left|\right|}_1$

Wherein, Φ is the bidimensional space-time code perception matrix that calculates of parameter by experiment, and ρ is the high-quality magnetic resonance image (MRI) of rebuilding out, S ' be smoothing processing after data, λ is the weighting factor of sparse transformation and fidelity item, and Ψ is sparse transformation matrices; Solve formula above by iterative algorithm and just can obtain high resolving power, high-quality magnetic resonance image (MRI) without aliasing artefacts.

9. as claimed in claim 1 based on spiral sampling and the method for reconstructing of the magnetic resonance imaging of space-time code single sweep, it is characterized in that in step 4) in, described spiral sampling gradient needs to carry out corresponding optimization according to different hardware conditions and experiment article profile, enable its greatest gradient amplitude and greatest gradient switching rate meet hardware requirement, and the sampling density of signal area can be improved.